You are on page 1of 36

C.P. No.

13
(12329)
A.R C. Technml

MINISTRY
AERONAUTICAL

OF SUPPLY
RESEARCH

CURRENT

The Viscid

COUNCIL

PAPERS

Flow

in a Narrow

of Air
Slot

G L. Shires

Crown CopyrIght Reserved

LONDON

, HIS MAJESTYS STATIONERY


1950

Price 2s 6d net

OFFICE

Report

mmorsndm NO. M.46.


December, 190.

The Vucld

Flav of Air

rn a Narrw

Slot

- by G.L. Shires.

The propartzes sf the viscid flow of ar in a rectangular


slot having
The results of'
a rndth large xn compar~on lath its depth are mnvestrgated.
varuw
tests cre found to verify
thaaretioal
and enpxrxal
relatlonshrps
between the pressure dxstrrbutlon
in the slot, the au qass flow and temperaEuth 1aruna.r and turbulent
flows are conture, and the slat dimensions.
s1dered.
CONTENTS
I .o

Intra&u3t~cn

2.0

The Equations
2.1
2.2

2.3

The ;ippwatus

4.0

The Aralysls

Coefficient

:
5

and the E'iethzd of Testug

af the Test Kisults

case 1. Constant Crass SEctxon


Case 2. Constant Depth and Incressmg
Case 3. Constant Xdth and Increzslng

7
ii
9

\i'ldth
Depth

Conclusions

5.0
List

Flw

The TheoretxxilEquat.tlon
for Lanruar Flow
The Empuxal3quatk~1
usu-~g the Resutvlce
Solutions for Thrct Partxular
Cases

3.0

::;
L.3

of Vlscld

10

of Illustrataons

IO

L3.st 9f Appendices

11

References

II

Appendices I to IV
Figures
1 .o

12

- 16

1 to 13
Introduction

The three foroes affecting


the flaw 'of au in a slot are the pressure
force, the force due to vlscoslty,
md the force requrred. ta accelerate
3r
deccelerate the flud.
If, however, the depth of the slot is very sm-11,
the vxad
fxce and the p~%~-e
force are very lar;t,
and In oonparlsan
the inertu
farce 1s small enough to be no&ected.
This bang so, the
viscid force 1s equated to the pressure fwce and a simple theoretxal
equation relatln&
the pressure distribution,
the ar temperature,
the nass
To
flow and the slot &rnensuxx is obtzuwd for steady 1amin.a.r flaw.
derive a sxular
cquatl3n for turbulent
flow the problem 1s considered in

Although the above analysis 1s concerned ,iith the pressure drop alow, ;
narrwcc rectangular
slot the exparmental
results may also be expressed in tams
of %ch number, !4. For thu purpose tke mean velocity
bzsad on IY=SS flew,
iii= - G , 1s taken as representative
and values of ii calculated from the
c.a.h
expression Id=--

-. G
p.a.h

mOver the test length the i,&.ch nmbor was al-mys low and vlriatlons
mre negligible.
Hensti, ovc:c this sectron,

of temperature

where To 1s the roscrvolr


tmpperature.
Typical curves of statw pressure, pa
and Mach number, Xi&,over the test length of a parallel
szded rectangular
slot
(case 1.) are shorn as full llnds 1x2Fig. 14(a).
In the last inch of slot tnerc 1s a large drop in pressure and density
and hence a rapid acceleration.
In 'mm length the varmtion
of teqerature
cannot be neglected.
If adlabatlc flmis
assw&
zn this part of the slot
the exit mch numbm, &, can be calculated for case 1, frow equations glvcn
by Kestin and 0ppenhem.W These can be rcducud to the expression:

where subscripts 1 and 2 denote two posltlons


separated by a distance L.
The reslstancc
coefficient
h 1s assumed to be the same all along the slot,
since the Reynolds Number 1s constant , and Its valut: I.S caloulated f'ron!
measurements in the test length.
The calculation
of the exrt Iwch number,
&, 1s in effect an extrapolotlon
of the ourves of Xach number in the test
length with the assumption of adlabatlc expons~on ln the last Inch.
Such
axtrapolatlon
is, of course, necessary since the prossure JUST u%?&e the
slot exit 1s not always ntrospheric.
Vhen the slot 1s choked it IX&Y be mu&
higher as in case (b) in Fig. 14(a).
Case (a) 1s unchd~ed.
When Me is plotted against the pass flow, Pig. 14(b), IC attains a
constant value, not necessarily
equal to unity, lndloatlng
tht choking
condltlons
were reached, l~:lth the four larger clearances.
The fact that
it, does not become unity nr.y be due to It bung based on the seen velocity
whereas the exit velocity
profllz
was probably not unifor;- and choking
?rabsbly occurred when the LJ~~XXILPvoloc:ty
reach-d the: local sonic value.
If this 1s true then the apparent Xach nubur, if,,
correspondlw
to choking
the exzt 1s effectlvcly
a profile
factor.
The reduction of ths choking value
of Pi, with decreasing clearance sows to indicate the truth of this assumption.
w J. Kestin, U.K. OppenheuL - "hi: Calculation
of Comprcsslble Flurd Flow by
thz Use of a Generalised Entropy Chart (Equations 17 and 65~1). - Inst. of
Nech. Eng., Proc. 1948, vol. 159, Tar "Iimgancy Issue No. 43.

terms of the resistance

ooeffx~ent,

h, a-13 the R:cynold's

number, Re.

The relationship
betweenh and. Rc for the flo;i of an mncofi\presslble
f'luld in pipes and slots has been a subjeot of research for marq yexs,
but
little
information
1s wadable
ooncern~-~g the flow &-a csmpresslble flud
subject to large char.ges in pressure and therefore to 1ar-r~ changes III Censrty.
Experiments have therefore been performed In slots of various shapes on the
flow of air subject to such changes, and the results,
-when expressed In terns
of the paramtters A and Re, are compared with the results
3f experiments with
incompressible
fluids carrIea gut by Ylasius (Ref. 1).
2.0

The Equations

of TIxxzd

Flow

The theory of the steady lsnunar flow of d comprcs;~ble v~scd flurd In


a slot the crabs section of whxh 1s constant, w changing wry slowly, may be
developed from first
prmnciplw,
tk r&hod bezng an extension of that used in
hydrodynamic theory with a function
of temperature and pressure replacing the
density term.
For turbulent flow use IS made of the concept of the resistan%
coeffv&ent,
A, where

T being the "skin friction"


per unit of surfzce xc& In oontlct slth the fluzd.,
The ralatisnship
< the spatial mean velocxty, and p the densrty of the f1uz.d.
between h and the Reynolds number, Re, based in the bydraulx
mean depth for
turbulent
flow 1s given enp~rxally
by Blas~~s (Ref. 1) as

and. this value can be used to develop sz~ equatlx


tax&
theoretwally
fsr la~nar
flow.

corresponding

to that

ob-

The assumptions made in the derlvatlon


of the equatwns of flaw are the
same in both cases.
The farce requxxd to accelerate or deccclerato
the
flud
1s assumed to be negl:llglble;
the pressure dxdrlbutlon
over s.ni cross
section is assumed to be unif'on,
and. the temperature of tht air as It flows
along the slst is assumed to bs constant.
2.1

The--____
TheoretxcalEatlon
--

Consder an elementary
at a point In the slot.

for Larmnar
_---- ---Flow
-volume of the fluid,

The inertia
tern b*lng neglect&,
element m the dlrectlon
of the x--s,
The vised
The pressure

force

= - g

which,

these forces

since s = I? . &
3Y

pressure,

i.'e obtan

anadS
ay

and
where s 1s the shearing

. 6X . 6;1 . 6e,

stress in the fluid,


r, 1s the absolute
the velocity
in the x direction.
By equating

sltuated

the resultxng forces nctwg on the


l"hich 1s the &xectxsn
of flow, are:-

: &Z . by . 6x . 6z
ay

fx-ce

as shown m Flg.la,

p 1s the denszty,

ancl u is

the relatlonshlp

&
a9

p.

, may be written

as

.l.....

dor the flow of a compressible


cent walls It 1s assumed that

fluid

-2-

III a sld

hamng two closely

..(I)

adja-

For the niz~row slot,

therefore,

cquatmn

iE?=,.i.s
8.x
Integrating

this

iilth

.*...**...*.

(2)

3Y2
to y we obtain

respect
u

(1) becomes

&

= L
2p

. y* + c, .Y + s:2

* a.x

The depth of the slot 1s h, v/here h 1s a fun&Ion of X, &,


when y = 0 or y pi h, we have
h .a
and c* = 0
Cl
=
--&
dx '

3me

U= 0

. . . . . ~ *. . . . . . . .

and this

1s a parabolx

The mean velocity

form of velocity

.(3)

profile.

at any CIDSS sectim


h
1
iy = -.
u@
h
j/ 0

1s

h
1

= h'ax

y2.aY

2.E.
s0

2li

= - 92 .A?ax 12 P

**..

. . . . . . . . . ...(4)

If the mdth of the slot be denoted by r, where a 1s a functmn


the mass flow along the slot LS then given by
G =

?.o.h.u
- 3 .".

&
dx
and, substituting

h3

12c

-12.v.G

', . c . d

m thzs equation
c&
ax

or 2 . dp

of X,

the vslue

;' : 1 , v<e obtam


RT

-121-1.RT.G

p.a.h3
=

.dx

-?2p.KT.G

If T, c', and h are known fun&mm


the theoretmal
pressure dmtrlbutlm

c . 0
of x, th-s equation
so obtaaad.

. . . . . . ~. . . . . . . (5)
can be mtegmted

and

the z direction
being umty.
When cons:dermg
ment the assumptions made are that:1)

The kmetlc

2)

&=a
ay

the forces

energy changas uv~~lvsd

acting

on this

ele-

are negllglble,

= 0.

dz

The force, F, rcslstmg


the motun of' the flud
may be consdared ns
due to the f'rlctior,
acting &t the surface of contact between the flud
and the
w.lls.
If the "skin frlctlon"
per unit area of "wetted surface" be denoted
by 7, then

I.e.

7 . (AE) . DOS 8 +

27.6x.

The pressure

force

acting

'u = p.

7 . (CD) . cos 6

m the &e&Ion

of motxon 1s given by
I (AB) sin @+ (C3,i an

h+(p+l&.&)

zax
- (> + ii? . 6x)(h + dh . 6~)
ax
ax

=p. h + (P + 5 2 . cx)
whxh

- iP + g

. &x)jh+

. 6~)

. . . . e,....

~ . . . . . . . . (6)

reduces to
P =

Equating

-h.&.fix

these two forces


2

7.

ax

F and P Tye have


6x

=-h.&.Sx
ax

&=
ax

or
Now T/$$

-- 27
h

1s the resdxmce

coeffux.ent,

h,

and hence

I_

c&.-=-h'k
,dx

* u
h

The coeffiolent,
A, 1s dependent on the Reynolds number, Re, apertainug.
For laminar flw the value ofh~ v&ah gives an equattlon da.tloal
t-ulth (5) in
section 2.1 1s

For turbulent

SubstltutughL

flow Q is given empiriully

equatitlon (6),

g=

-2.

as

we have

-,=
h

where

by Blasius

-3.
Re

G2

,, .c 2 .h 3

or
whxh

= - 12 p.RT.G.dx
c . d

P . dP

IS the same as equation

(5).

To obtarn the correspondug


rn equation (6) the value of $.

equatxon f3r turbulent

+p=-~.~=
Rei

-0.079.
1

ReL

flow we substitute

G2
p . a2 . d

7
= - I0 067

i;
G

pt.
I

h3.

a4

.p
I!

or

p . ap

p . RT .

= _ 0.067

a;:
.

2.3

G4

. &

. . . . . . . . . . . . . . ...(7)

Solutions
--

fpr Three Partz~cular


.--"-----

Cases

The general equatrons, (5) ad (7), can be integrated


lf o, h, and T
are kncwn as functions
of x.
For the three types of slat dlustrated
in
Fig. 2 the mtegratmns
have been performed on the asswmptlon that the
temperature,
T, 1s cxxtant
along the slot.
Case I,
--result

a mnd
----.h oonstant
---_

The lntegratlon
of equation
for steadjr lamu~~r flow.
p2

or Pq

=.Z~IJ

- P22

gives in this

(5)

case the fdllowing


x

.RT

. G(+

-x1)

. . . . . . . . . . . .(3)

. . . . ..*......

(5)

0 . h3
On lntegratlng

equation
73

we obtain

(7),

ECconstant

for turbulent

flow
. . . . . . . . (IO)

Case 2. h constant
-_-_
----.-p--f-_ and a= cx x
If
-in this
circular

the frIctiona
effect of the two side walls 1s negleoted, the flow
case 1s analogous to the r&al
flow between two wLfx%Ly
spaced
flat plates and by symmetry the pressure at any radius 1s cons'csxt.
-5-

ProvxM.,
therefore,
that the tivergence 1s small the pressure in any plane
perpendmular
to the axis 1s approxmately
constant and equatmns (5) and (7)
may be applied.
By substituting
in these equstlons & = o( .s, w obtazn for
larmnar flow
p,ap
whxh when integrated

= constant

or PI 2 - P22

ap

P.

- 24 u .RT . G . lo&
q . h3
i

= 24 u . RT . G . log,
o(. h3

x2
T-

. . . . . . . . . . (12)

>

. ..*......*.......

(13)

fla7,

I
0.067
y:
.
RT
.
G'
1
ct.4 . h3

which when integrated

becomes

= constant

p2

e
x

becomes

P2

and., for turbulent

-12p.RT.G.
cx . h3

kdth

&
1
x4

I
0.178 P . RP . G4 . 1
I
x4 i

z
GC4 .h3

Case 3. Constant

and. Inoreaslw

. . . . . . . . ..a.. (14)

Depth

As the deviation 3f the two plane surfaces is very small, the components
of velocity
perpendicular
to tht ax~ of the slot must also be very small, and
the pressure distribution
zver any cross sectlsn x,therefore,
assumed to be
dorm.
For laminar
P *

dp

flsw,
_

and %ghenintegrated
p*

or PI 2 -p22

by substituting

121-1.RT.G
@ .P 3

vkch

constant

we abtam

&
x3

+ '2 p 6 RT * G . 1
i \ a.!3 3
x2

12i-1.W.G

. . . . . . . . . . . . . . . . . ..(16)

. . . . . . . ..*............
(7) we have, for

turbulent

(t7)

flow,

- 9.067 p' . RT , Gt . +
3
1
3
04
* p

when integrated
p*

(5),

thus becomes

3
a - B
By the same substitution
in equation
P . dp =

h = p . x in equation

gives

,..*,...*.*...*

-6-

(18)

.RT.G
or

PI

p*2

0.067 I cl;
04 .

8,

1
x1
11

A- xz2 I

119)

.*............

p3

of two fl2t
The a~p~aratus, as shm,n m Fly. 3, consisted essentially
steal plst'3s sepwated by thin strips sf st2e1 arranged so as t3 form a
The two plates uutire made I" thick to prevent
passage of the required. shape.
8
any appreciable bchdlng, ana, after the upper one hd b?en case h~dencd,
ground
the surfaces forming the hvrleonta.1 iinlls of the slot were carefully
The m~.~mnundenatum
fron the plane,
and tested against an optical "flat".
as estimated by cxntlng
the lnterfzrenoe
fringes,
was about 50 , 10-6 inches.
Between the tws plates the sl;t *,ns bouded Qn each 31de by strips of steel,
whxh In the case 3f the slots sf constant depth wre of single thickness
but which far slots of vnryzng depth were overlapped In the form 3f steps.
i;'hen the blzck was nsstmbled and firmljr clamped togtther
a szallng conpoUd
was applld
uxternelly
ta -11 the jomnts.
nlr nt z recorded tempedxrre,
T, passed
During the t&.sts cx-presxd
rtservoir
1n
through a wtcrlng
scctlon and lnt- t'he sharp edged rectangular
From there It flowd
along the rectangular
passaga between
the upper plate.
pressure being recorded at intervals
the two plates tc atmosphere, the stdx
A line of 7 grLssure tlpplngs
across the slot at
of :I" docg ths 1x1s.
right angles to the 2x1s was use& 2s a check upon the setting of the block,
since any ;nclir,atlon
rf the tv3 surfaces In a plane perpendicular
to the
axis would rtsult
in 2~1~ssyrrwtrxs.1
pressure tistrlbutlon
over these 7
pslnts.
The pressure dlstrxbutxn
at this cross sectron was found to be
umforn as nssuned m the thecry.
The procedure for a<ach test -da3 the same and cznsrsted of two parts.
Th6 first
%ras the recwding
of the xlal
pressure dxtribution
for VwlOus
It
values zf the reservoz
pressure, po, and the anjlysrs
sf these results.
was lnverxbly
found that the prxs-sure dxtrlbution
avtir the central 3"
length sf slat cudxmtd
t3 thz correspstilng
theoretical
form and thd this
The second part of
length could be cx~s~Lred as a sultsble ta:st length.
the procedure involved ,-n xwzstlgatlon
of the rilat.tlxxhip
between the a~
mass fl<w, G, in the slot and the prtssures p, and p2, where p, denotes the
pressure at the upstream end <andp2 the pr-essure at the downstrean end of
the test length.
4.0

The hnalysls

of the Test Rcsdts

The aim of thz tc;ts r>as to check the valdity


of the formulae dzvel0ped
u? sectrons 2.1 and 2.2, and, T-here pzsslble,
to aetermlne tht relationship
betmew the reslstnnce cdtfficlent,
A, and the Reyngld's number, Re.
SlnGC
the cquati,,ns Involve both thz wzdth, a, and the depth, h, as varrables,
It
was necessm~y '1s an overall check t3 use thrte types of sld,
one with il
constant crass se&Ion,
one with a varying width, and. one rslth a varying
depth.
Equations (8) to (19) a;j-ply to the thrcu types of slst used and
are coripared in thi f'Jlc~J,mg pages with the test results
obtained.
L.1

Case 1. Constmt

Cross Section
-

Tests wre 'wdde with six slots havlng 2 w&h,


n, of 1.75"
h, of approximately
O.OOZO", C.OOjO", 0.3040", 0.0050", 0.0058",

and a depth,
and O.OC93".

Fig. 4 shws the axial pressure tistrdutlons


as recorded for two 2
the slots, the dlstrlbutlons
obtalnzd with the uthw four slots bexng of
the same form.
ilhen the square of the absolute pressure, ~2, is plotted
lines are obtained. aver
agam.st the axial drstance, x, a series 3f straight
the central 3 ' length of the slot, the f'wm of the theoretxal
rcls.t.tlsnshlp
bstween p znd x bt.lng thus vcnfxd.
The slope
of the rcslstance
POmhI& mnss flm,

df each of these straight

lines ,(PLz 1 "," j

, is a measure

to flaw and is plstted l~garlthdcall~ngalnst the OOFXSHere, the SIX curves represent the results
G, in Fig. 5.
-7-

far the six slots investi&trd


and, SUT.X mtslde the transltlnn
z3nt each
set of results oar, be represented by two straight
lines, the relationship
between the pressure drep and the mass flow is c&arly
of the form
- P22

PI2

= Constant

G" .

- X1

X2

The constant in this equation is different


for laminar and for turbulent
flow
as is also the index n, which for larunar flow is 1.0 and for turbulent
approximately
1.75.
The non-linear
part of the curve represents the transition from laminar to turbulent
flow.
The experimental results
s.r~ also shown as pJint,s in Fig. 6, together
with the theoretical
curves obtained by substituting
in equations (8) and. (Ii)
the slot dimensions and the air temptirature.
The difference
b&wean the
corresponding calculated
and measured values 3f [P12 - p22j as a percentage
\

x2

- x,

of the measured value, is with one exccptron less than $ for 1armns.r flow and
less t hen 1% for turbulent
flow.
The larger overall discrepancy in the case
where h = O.OOjC'" is probably due to a small error in the measurement ,~f h.
To complete the picture a graph is given in Pig. 7 of the resistance
coefficient,
A, against the Rsynold's number, Ra, where h and Re are calculated
according to the equations derived in Appendix II.
For laminnr flow the
points lie on the theoretical
curv,,, XT> = ?L!L, but for turbulent
flor-J the
lie
0.079 , as stc-hod by Blasius gives values of the
empirical
expression, hT = ._
Em;
resistance coefficient
Lswr than those calculated
from the test results.
The experimental points for turbulent
flow lie close to tho two curves,
hTsyandhTs?ip,
th,: first
fitting
the results for h = 0.0033" and
Rez
ReZ
the second the results for h = 0.0058" snd h = 3.0040".
Except for the case
where h = 0.0050", the results indicate that the smalltr the value of h the
closer lie the points to the e?prricni
ourvz,hT
=q
.
ReT
The tranmtion
from laminar to turbulent
flow takes place over a range
of Re wh-uhichis slightly
different
for each slot.
These ranges are given in
Appendix III,
the average values of Re at the beginning and at the end of
transition
being 2,120 md 3,810.
4.2

Case 2. Constant

Depth end Increasing

?idth

Two slots were used, each having n constant depth, h, and an increasing
width given by n = (;( . x.
The drpth in each case was O.C036", and the values
of awere 0.1 and. 0.2.
reveals

The axial pressure distributions


that these are of the form
p2

p2

= C+D.i

are shown in Fig.

A-B.log,x

The analysis

8.

f3r lcaminar flow

and
for turbulent

flow.

2
A, a, C, and D are constants,
and x is the axial distance from the imaginary
pslnt of intersection
of the two side walls.
In this ease the analysis is
nat SJ accurate as for case 1 and, although a strazghht line is obtained by
plotting
p2 against x-0.75,
this figure cannot be taken as exact since a sin=lzr straight
line may be obtained by plotting
p2 against x", where n rs.ng-3~
fron 0.70 to 0.80.
The slight scatter due to errors in the rea&ngs of p is
sufficient
to mask the small deviations
from the linear obtained with the variation of n over this range.
In the first

G" of slot

length

there was an appreciable


-8-

divergence

from

and a central 3 length was


the thexotxal
fern of pr ussure distributi-n
A 5631'1es3f rsd.1ngs of p,, p2 end.
^ G
therfore
chosen as the test length.
the pressures at
were taken for each z&t, whore p, and p2 :>ers rtspLotively
The values of
the upstraar. and at the dswnstrear, end of the test length.
(~1~ - ~2~) are plotttd
l~garrthr?lcelly
ag3.mst the corresponding values of'
straight
lmcs,
G in Fig. 9 and the results are sax tz lie m tv3 inclined
thus indicating
the relationship
p,2

- gz2

= Can&ant

. G;"

The index n is 1.0 for larlnar


flaw, but ftir turbulent flow it dlf'fers in the
Alss shorn in
tw slots, being 1.85 vihc:n(r( = 0.1 and 1.60 xhhenrx = 0.2.
Fig. 3 arc thi: corresponding thesretical
curves obtained frsn equat.tlons (13)
The thaor&xal
curves for lapnnar flow are ldentxal
>ath the
aId (15).
teat results.
The Reynolds nunbcr, Re, being a function of the ,ndth, c, varies along
the slot for any given value of G and it 1s not possible,
therefore,
to express thr? results as a graph of h against RL.
For the sarxe reason, the
change from lanunar to turbulent
flow does not take place at the sar~1evalue
of G in both slots, but is determined instead by the conditions
at the entry
where the Keyn~1d.s nunbr;r is highest.
The valws of Re at the entry correspondlng ta the beginning of the transitian
are 2,760 <and.3,020 form = 0.1
andc:(= 0.2 respectively.
Tht critxal
value for a parallel
slot of the
sane depth is estlrlattd
frori the values in Appendix III as 2,400.
4.3 Case 3. Constant %',rdth and Incrznsing

Depth

Two slots of constant width equal to 1.8 ' were used, each having a
depth given by h = fi . x,xhere' e was 0.5 . IO- 3 in one test and. 0.67 . IO-3
1.n the Jther.
central

u.hich when analysed over the

FFg. 10 shoxs the pressure tistributxns,


length nre found t3 be of the fx~
p2

= x+02

for both lxuner


and turbuL.nt fl>w, where E and F are constants and x is
the axial dastvlce frdjn the ~xagmary line of intersection
of the t--o planes.
As in the other cases, there is a considerablz
divergence fron the thexetxal
relationship
in the first $', of the slot and a contra1 3" length was again
chosen as the test section.
Readings of pi, p2, and G were taken and these
are plot&d
logarlthrncally
m Fig. 11 as (p ,2 _ p22) against G.
As in the
previxs
oases the rtilationship
between the pressure drop md the nass flow
1s of thr3 fxn
i-q2 - P*2
The index n is 1.0 for

= Constant

1,mma.r flow

. Gn

and 1.75 for turbulent

flaw.

The theoretical
Landempiracal ourvos obtained from equations (17) and
are also shown ~.n Fig. 11.
The d.xwre~anoy between the test results
and the theoretIca
curves fur laxlnar flow is unexpected., since all previous
lsmumr results have been in close agreement snth the theory.
For this
reason an error in the calculation
of e was suspected and it is shown in
Appendix IV that a difference
of la;?. in the theoretrcal
value of (p,2 _ ~22)
rwy be duo to a difference
In lj of less than 49.
The discrepency rosy,
therefore,
be due to an error in the measurement 3f h of onlyQ.C0J2", an
error which IS quite possible wath the stopped shims used in this case.
The
laminar rssults 'iere accordingly
assumed to be correct and new values of p
calculated.
(19)

Using the correotzd values of @, hT candRt were calculated


from the
equatxons in Appendix III and arc plstt-d
in FF~.
The turbulent
results
1~ on ix3 fumes, AT = q
and AT = &&
:2ior p = 0.64 x iO-3 and
ReS
Rex
-9

c = 0.49
x 10-3 respa3avely.
It appezrs that the constant (XT , Rei) nay
be dependent upm the sld divergence and sz It 1s plAted
agamst I: m
Since
the
nean
&pth
3f
the
tw
slots
used.,m
this
case was less
Q.
13.
the value of (AT . R&) cxrdspmding
to
than 0.006" (see section 4.l.j,
e = 0 (Case 1.) wz.a taken as 0.081.
F1.g:. 13 shms that the three pomts
lie apprximtely
m a straight
lme, thus mdmattlng
a possible relatmnship bttweenhT,
Re, and [ of the fxm

hT = 0.81

- 28 p
Rei

However, further
5.0

tests

ml1

be necessary

to vbrd'y

thus hypothesis.

Conclusions

The cmparison
3f the ti-st results
and empmical relatlonshlps
shows that
fx the flm 3f air ux ,md nammv alot,
and that the length consderod do,us mt
x the slzt txit.
\:lth these prxisims
kinetic
energy are negligible.
so that
thexy
1s valid.

vxth the correspmcling


theoretxal
these relatlonshrps
hold apprsximately
pmvlded that the &vergence 1s mall
lx wlthm about 2" 3f the slzt entry
the fxces
&ue ta the changes in
the r.ssumpt~sn mde m developmg the

The results also mdicatz


that thi fhcxeticsl
equations my be used
t3 give an ,mcurate qumtltatlve
discrlptlon
uf lamnar flaw.
The only
serious discrepancy bitwen
the themy md the test results was in the case
3f the slots with mcrdasing
d.tpth rind here the error was assumed to be m
the masurement of the dlv&rgmct.
sOpxd
Fx turbulent
flow th= enplrical
rilatlonshl
m the value of
B r , are nd m general
the msistance
crxfflclent
g~vm sy alnsms, hT = A
R&
applicable.
The investlgatlm
shows that for slots of c,nstmt
cross
siction the actual value ofhT 1s lnrger, bemg 0.087 for the slot 3f depth
Ro+
h = 0 . 0093" and 2-q
for h = C* 0058" .
Far s1As 2f xxcreaslng depth the
,ReZ
product (AT . Re") appears tu be z constant, thL value Llf whxh depends upon
the rate af divergence, fi, zf the mlls.
Llbt
V~scld

3f Illustrntlms

Flsv Thedry.

The Tj-:>es of SlA

Cmsdered.

The T'dst Black.


Fressurc
Effect

Dmtrdmtlon

m the Slot

3f Pressure Drop on Mass Flm

Conparmon with
Scde Effect

Theory.

(Case 1).

af Pressure

m the Slst

(Case 2).

i~rap on siass Flow (Case 2).

Pressure Dlstrlbutlon
Effect

(Case I).

in Case 1.

Pressure Dlstrlbution
Effect

(Case 1).

=n the Slot

(Case 3).

of Pressure Drop anMass Flow (Case 3).

Scale Effect
The Effect

m Case 3.
of Slot Dlvur~ence.

Mach Number Varratlon


- 10 -

ik!lsranaum

List

No.

M.46.

af Append~es

Appendix I

Lrst of Synbols Used.

Appendix II

The Calculatrsn

Appendix III

Trarmtion

Values of Re.

Appen&x IV

The Effect

of Ermrs in the Neasurement of @.

sf h and Re.

Refzrcnces
Jg.
1.

Author
N.A.V. Piercy

Title
Aerodynamics:

-11

pp.

278;

E.U.P.:

1937.

mm1x
List

3f SyriDols Used

absolute

pressure

aens1ty

zbsolute

!J

coefficient

af absolute

ooefflclent

of kmemtic

width

depth of slot

I.

length

Ilass flaw

valne

ii

man velocity

surface

reslstanm

Re

Reyndds

tenperature
viscosity
vmcoslty

of slot

of test

sectmn

I?f flow
=

-Qc . h

frxtmn
T
= p ii2
2

coeffloxnt

number

2 .h

.u
11

- 12 -

AP?ENDIX II

The odculztlun
based on the formula

of the Rejwdds:

Re
where

nuiioer, Re, frzil

results

1s

= 4m.c
v

m =

the mean hydraulx

ii

the wan velncity,

and. v

the klnematw

Fx a slot af rectangular
ble compared with a, the width,

Therefore

the test

Re

depth,

vlwx'slty.

cross section

where h, the depth,

is negl~gl-

= =.h.ii
P

That is

equatix

whxh,

Re

= ?s-.
fi . P

Fran oonsderdlx
(6) 1s dw~ved

since

and, replacing
p.ap

. . . . . ..*....*..*m

af on ele;ler;tary
irk .sotmn 2.2.

p L Li , a . h ,

fip

=-A

dx
. G2
J . n 2 . h3

;l by $

volume of air
This g~vos

(20)

between the two plates

bccomzs

, : e have
=

h . RT ~ G2

. dx
. . . . . . ....-.......

c2 h3

t 21)

kssum~ny that h can be expressed as a function


of Re, whuh 1s udepcndeat of x when CI is constant, and that the temperature,
T, LS constant this
equation can be u&grated
to give
2h

P,2-P$

. R'I' . G2 . (x2 - x,)

=
a2 . d

- 13 -

ma hence
h

PI

2 - P22

c2 h3
. . . . . . . . . . . ..(22)

2 RT . G2

x2 - 7

For case 3, zn which the aepth of the sL3t 1~ zncreasing, we cm obtan


a carrespondmg equation for A by substduting
h = CCCin equation (21) and
by integratmg.

ana

This is p , dp

P12 - P*2

- A.

RT , G2 . &
x3
2 63

A.

RT . G2 (1
a2 p3

2
Xl

-2

2)
x2

....o..,,,.......*

- II+ -

(23)

Ax?ENDIX III
Transltmn

flm

The range of m.sa flm


takes place 1s slightly

The followmg
flm far each test,
fron the relation

V'alucs of RC

over whmh trans~tmn


fro?; lznunar
ddferent
far each slzt.

table gives the approxlnate transition


range af nnss
tqether
mth the correspon?ang range of Re czlculated

Range of G
lbs ./set.
(22.9

(20.0

- 28.8)

(16.6

- 37.2)

(17.38
(15.9

to turbulent

lb/ft.

- 31.6)

Range d' Re

1.189

. lO-5

2,640

. 10'4

1.198

. 15-5

2,290

- 3,290

, IO-4

1.201

. 10-5

1,890

- 4,250

1.198

. lo-5

1,990

- 4,150

1.225

. IO-5

I.780

- 3,540

) . 10-4

- 36.3)

sec.

. 1O-4
, IO-4

-/

The cean values of Ke for the beginning


2,120 and 3,810 respectively.

m-13the end of transrtion

%ith
th e available
au supply it was not possible t'3 obtam
in the slot having a depth of 0.0030" and the higher limits
cannot,
therefore,
be given XI this case.

-15

are

turbulent
flow
of transltzon

APTNDIX Iv
The
--- Effect

of Errxs

m the Measurement of 8

Fig. Ijb shows n se&Ion through a slot of ~ncreusm& depth, shomng


h' end h" the clearances vvhxh aefme the dlvergenoe, @. The depth, h, ~3s
assumed to be equel tu the thmkness of the steel shin at the pomt consdered
as measured by a micrometer.
When p2 1s plA3xd

agamst

a straight

of x are therefore assumed. t3 be correct.


errors vJhhlch ocmm being m the oalculatim
The theoretxal

ti

empmxal

line

IS obtund

and the values

The pod
0 is then flxed,
of the angle GOA.

equstims

for

slots

the

3f th1.s Hnd are

PI2 - P22 = 12.!~.-.2T


0 . p3

pq2-p22

and

0.067 ,I -; . ET . G4
04 3

for turbulent
flow,

--3

An error m the measurement of h gives a false value


it occurs in the equations at f3, gives a very large error
value of (pq2 - Pan).
Consider

the case Tfheherethe measured values


h'

2.0".

of h cre
h"

lo-3

of @ whxh, since
m the theoretxal

Then

6.0".

lO-3

0.667 . 103
Suppose that the actual
h'

Then P*(the

real

value)

=
=

values

of h are

1.933.
h" - h'
x1' - x'

10-3

=J.667.
6

The resulting
error in (p,2 - p,2)
a maximum error, Sh", of only 0.0002".

h"
10-3

IS, therefore,

- 16 -

5.8".
- 0.6L5

lo-3
.

IC-3

11% and. LS due to

FIG. I.

VISCID

(4

ELEMENTARY

FLOW

THEORY.

VOLUME

OF FLUID

(b)

ELEMENTARY

VOLUME

BfTWEEN

TWO PLATES

FIG. 2.
THE

TYPES

OF SLOT

CONSIDERED.

--

(4

CASE

CONSTANT

CROSS

(L = CONSTANT

SECTION

a=d

A=
CASE 2

(b>

IN CREASI NG

CONSTANT

WIDTH

Q = CONSTANT

k--k
03

CASE

INCREASING

rL

DfpTH

= p

FIG.3.
THE

TEST

BLOCK.

FIG 4.
PRESSURE
(CASE

DISTRIBUTION

I : CONSTANT

CROSS

IN THE
SECTION,

SLOT

I .75)

a. =

60

01

Ccl)

I
I

DLSTANCE; FROM RRLRVOIR


DEPTH -R = 0~0020

DISTANCE

(4

I
2'

DfPlH-R

FROM m=VUlR
- 0amo

INCHfZS

INCHfS

43

FIG.5.
EFFECT

OF

PRESSURE

I : CONSTANT

CROSS

DROP

ON

SECTION,

FLOW.

MASS

a = 1 .75)

TURBULENT
FLOW.

_.L

TRANSIl

3.5

3
IN

t.0

FIG. 6.
COMPARISON
(CASEI :

CONSTANT

WITH

THEORY.

CROSS SECTION, a = b75)

3.5 /

//

FIG. 7.

SCALE

EFFECT

IN CASE

I.

0038

0 034

0 030
I

!2
w

OGO8

lO,QQQ

5,QOO

RCfNOLDS

NUMBER

Re

FlG.8.
PRESSURE

DISTRIBUTION

(CASE 2 : CONSTANT

IN THE

SLOT:

DEPTH , 4 = 3.6 x iOr3a = o< SW)

60

I
0

I
DISTANCE FROM
WIDTH

INCREASINL;

,a
I

DISTANCE
WIDTH

o(

RfSfRVOlR

INCREASINq

43l

= 0-I

2
FROM

4
INCHES

2
3
RESERVOIR
-

M=OZ

INCHES

49

FIG. 9.

EFFECT

OF PRESSURE

CASE

2 :

CONSTANT

DROP ON MASS FLOW.

DEPTH,

38

= 34.5

IO-; Q=d*Y

LAMINAF
FLOW

THEO:OREITICAL RELATIONSHIPS
SHOWN AS BROKEN
LINES.

z.8

LOG10 G - G

3.0

IN

LB./SEC.

52

34

PRESSURE

DISTRIBUTION

( CASE: 3 : CONSTANT

IN

WIDTH,

30

THE
I a3;

DISTANCE FROM RESERVOIR


INCREWNG,

(9

DEPTH

90

/3.x)

INCHES
= 0.5 x la 3

80
70

--IL =

FIG IO.
SLOT.

TEST

LENCTH

y=

3*

60
50

40

30

20
IO
0
0

DISTANCE

INCRE,WN~

FROM

2.

RESERVOIR

DEPTH

- p = 0*67*X

INCHES
lO-3

4-9

FIG. II.
EFFECT

OF
(CASE

PRESSURE
3:

CONSTANT

DROP
WIDTH,

ON MASS

a = 1.83:

di = p .x)

38

36

THEORETICAL
SHOWN AS

T.6

5.0
LOG10 4

RELATIONSHIPS
LINES

BROKfN

32
IN

LB/SEC

34

56

FLOW.

FIG. 12.
SCALE

EFFECT

(CORRECTED

VALUES

IN CASE
0~

3.

R>

0 08
RESULTS
0 07

Q
A

006

,A=

(A=
/WC)
p = 049 x 10-3
p = 0.64

x ro-3

Rf?
0 05

0 04

0.03

r
SI
t

o-0;

r!?i
A=

0 01

0
IO

1000
REYNOLDS

2000

3000

NUMBER

Re

FIG 13
THE EFFECT OF SLOT

DIVERGENCE.

0
SLOT
(4

DlVER~fNCE

VARIATION

Of (h.

,f3 X IO3

Reg)

WITH

/3

0
B

THE: STEPPED

(b)

THE fFFfCT

OF ERRORS

IN THf

SHIMS.

MEASLlRfMf

NT OF /3.

f IG 14.
MACH

NUMBER

VARIATION.

DISTANCE FROM RESERVOIR - INCk


VARIATION ALONG THE SLOT, h = 5.80 x IO+ IN.

0a

+
/I

043 -

1
%
MASS &V

(9

EXIT

MACH

O-6
0.4
PARAMETER, $ANUMBER

ID
03
LB/SEC. IN!
DS 47612llIR.72

Kj

8150

DL

C.P. No.13
(12329)
ARC. Technical

S.Q.

Code

No

Report

23.90%.13